Lessons Learned from Hepatitis C Virus Next Generation

Sequencing Data Analysis

Eric Donaldson, Ph.D. Clinical Virology Reviewer

FDA/CDER/OND/OAP/Division of Antiviral Products

The Forum for Collaborative HIV Research NGS Symposium

March 12, 2015

1

Disclaimer My comments are an informal communication and represent my own best judgment. These comments do not represent FDA policy and do not bind or obligate FDA.

2

Cutting Edge Technology Ultra High Throughput + Lower Cost = Broader Applications

NGS Submissions at FDA CFSAN

• Identification of pathogens causing foodborne outbreaks

• Characterization of microbial contamination of the food supply

• Pathogen risk assessment

CVM

• Whole genome sequencing to monitor antibiotic resistance

• DNA barcoding for identification of foodborne pathogens

• Biomarker discovery of inflammation in food animals

CBER

• Vaccine safety

• Identification of vaccine contaminants

• Validation of virus challenge stocks

• NGS for HIV genotyping, tropism and drug resistance

CDRH

• General oversight approaches for NGS-based tests – germline and somatic applications

• Enhanced microbial sequence quality • Human and microbial reference

material selection

• Development of universal databases of clinically relevant variants

NGS Activities In CDER

5

Current uses

• Viral resistance analysis – Drug durability issues

• Viral tropism analysis – Drug feasibility issues

• Detection of adventitious agents in viral isolates

• Metagenomics of microbial flora of the gut

• microRNA profiling in pancreatitis

Potential uses

• Personalized medicine – Individualized cancer treatment

• Biomarker discovery (GWAS)

• Identification of novel drug targets, e.g., oncology

• Identifying genetic susceptibilities to drug related adverse events

Viral Resistance as a Biomarker

6

• Hepatitis C virus – HCV GT1a with viral protease Q80K polymorphism does not respond to

simeprevir – R155K or D168E in the HCV protease knocks out the whole class of HCV

protease inhibitors impacting subsequent treatment options – L159F or C316N in the HCV polymerase at baseline may be predictive of

sofosbuvir failure

• Human immunodeficiency virus – Baseline resistance profile of treatment-experienced subjects infected with

HIV-1 who failed HAART

• Influenza A virus – H275Y substitution in the H1N1 influenza A virus neuraminidase confers

oseltamivir resistance

• Identifying resistance-associated substitutions and baseline resistance polymorphisms is essential for determining the safety and efficacy of a drug and for identifying subgroups of patients who will benefit the most from using it

DAVP Review of Viral Resistance Data

7

• High level summary data in a format that shows polymorphic and treatment emergent amino acid substitutions (blank cells for insignificant changes)

• Tables listing each amino acid position with changes in decreasing frequency of occurrences in virologic failure subjects

• Information about assays and programs used for analyses and the parameter settings used for analysis

• Raw data for independent FDA analysis – Determine if treatment emergent amino acid substitutions can be correlated

with treatment failure (emergence of resistance) – Determine if any baseline polymorphisms lead to reduced efficacy – Determine if resistance-associated substitutions confer cross resistance to

other drugs against the virus

• NGS data adds complexity to resistance analysis

8

Importance of NGS

“Next generation sequencing has become a commodity.” ― Vicky Schneider

Why Sponsors Like NGS

9

Method

Single-molecule real-time

sequencing (Pacific Bio)

Ion semiconductor (Ion Torrent sequencing)

Pyrosequencing (454)

Sequencing by synthesis (Illumina)

Sequencing by ligation (SOLiD

sequencing)

Chain termination (Sanger

sequencing)

Read length

5,000 bp average; maximum read length ~22,000

bases

200 bp 700 bp 50 to 250 bp 50+35 or 50+50 bp 400 to 900 bp

Accuracy

99.999% consensus

accuracy; 87% single-read accuracy

98% 99.90% 98% 99.90% 99.90%

Reads per run50,000 per SMRT

cell, or ~400 megabases

up to 5 million 1 million up to 3 billion 1.2 to 1.4 billion N/A

Time per run 30 minutes to 2 hours 2 hours 24 hours

1 to 10 days, depending upon sequencer and specified read

length

1 to 2 weeks 20 minutes to 3 hours

Cost per 1 million bp $0.75-$1.50 $1 $10 $0.05 to $0.15 $0.13 $2,400

Advantages

Longest read length. Fast.

Detects 4mC, 5mC, 6mA.

Less expensive equipment. Fast.

Long read size. Fast.

Potential for high sequence yield, depending upon sequencer model

and desired application.

Low cost per base.Long individual reads. Useful for

many applications.

Disadvantages

Moderate throughput.

Equipment can be very expensive.

Homopolymer errors.

Runs are expensive.

Homopolymer errors.

Equipment can be very expensive.

Slower than other methods.

More expensive and impractical for

larger sequencing projects.

Derived from BMC Genomics. 2012 Jul 24;13:341 and J Biomed Biotechnol. 2012:251364.

NGS Provides a Deeper Look

10 Chevaliez, et al. 2012. Gastroenterology, Vol. 142, No. 6.

Analysis of a Breakthrough (Subject treated with an HCV protease inhibitor)

11 EOT = End of treatment

Observing Resistance Development (Development of resistance to an HCV protease inhibitor)

12

• NGS allows observation of amino acid substitution frequencies over time

Assessing Resistance Persistence (Persistence of resistance to an HCV protease inhibitor)

13

14

The Technology Challenge

“There are no facts, only interpretations.” ― Friedrich Nietzsche

NGS Data Complexities

15

Variant Detection

16

• Different callers were designed for different purposes

• Very few standardized methods for SNP calling • Most methods: remove duplicates and filter bad reads

• Little overlap among the different callers Figure from CLC Genomics Workbench Manual

Low Concordance SNP Detection

17 O’Rawe et al. Genome Medicine 2013, 5:28

18

Review of NGS Data

“The truth is rarely pure and never simple.” ― Oscar Wilde

The FDA Technology Challenge

19

• Industry adoption of NGS technology outpaced development of FDA review technologies

• Two high profile NDAs Spring 2013 for the treatment of Hepatitis C virus • Simeprevir (NS3/4A protease inhibitor) and sofosbuvir (NS5B nucleoside

analog polymerase inhibitor) • Both submitted NGS data for viral resistance • Both were discussed by Public Advisory Committees • Both NDA reviews were completed and the drugs approved by DAVP

• At the time of submission, CDER had no capability to adequately review/analyze these data

• At the time of submission there was no “business process” for transferring and storing the data

Regulatory Issues Pertaining to NGS Data for Drug Approval

20

• Experimental design: What exactly is being sequenced? Is the target amplified? Is it multiplexed? Are minor variants preserved? Are there experimental artifacts? Are there template issues?

• NGS data analysis: How are the data being analyzed? Are the results robust? What programs and parameters are being used?

• Regulatory review: What data are necessary to make a regulatory decision? Are summary data from one analysis pipeline sufficient?

• Data transfer and storage: If raw data are requested, how are these data transferred to the agency? How are these data stored long term? Is encryption required? How is data integrity managed?

• Data confidentiality: How are we protecting proprietary information?

Review of Sofosbuvir Resistance Data

21

• Determine if any HCV baseline polymorphisms lead to reduced efficacy

• Determine if treatment emergent amino acid substitutions can be correlated with treatment failure (emergence of resistance)

• Determine if resistance-associated substitutions of SOF confer cross resistance to other HCV drugs

No Resistance Detected Claim

22

• This abstract was presented at the American Association for the Study of Liver Diseases meeting in November of 2013

• This claim was based largely on the fact that S282T was a resistance-associated substitution identified in cell culture that rarely emerged in the clinical trials

Treatment-Emergent Substitutions in Phase 3 Trials

23 *All highly conserved positions

◄ PCR artifact

• L159F was present in 13 subjects who failed – Known resistance substitution of HCV nucleotide inhibitors (with L320F) – Emerged in 6 GT3a subjects, 2 GT1a subjects, 1 GT2b subject and present

at baseline in 4 GT1b post-transplant relapsers

• E341D emerged in 13 subjects who failed – Emerged in GT3a relapsers

• C316N/H/F present at baseline in 10 subjects who failed – All 10 were infected with HCV GT1b

• V321A emerged in 5 subjects who failed – Emerged in GT3a relapsers

• S282T or S282R emerged in 2 subjects who failed – S282T: GT2b relapser (12-week SOF monotherapy) – S282R+L320F: GT1a non-responder (SOF/RBV )

NS5B Polymerase 3D Structure

24

NS5B Polymerase Active Site

25

SOF Resistance Substitutions

26

NGS Confirms Treatment-Emergence

27

Conclusions

28

• Analysis of raw NGS sequence data allowed DAVP to observe: – Low frequency treatment-emergent substitutions of unknown clinical

significance – Cross-contamination of clinical trial samples – RT-PCR artifacts that confounded interpretation of NGS resistance data

• NGS may prove useful in identifying minor populations of resistant viruses present at baseline that could lead to treatment failure if the wrong regimen is prescribed

• Structural bioinformatics provided insight into the mechanism of resistance for substitutions that arose during treatment with SOF

• Access to NGS data and analysis software allowed CDER reviewers to make better informed decisions during review

Acknowledgments

29

CDER/OND/OAP/Division of Antiviral Products Debra Birnkrant, M.D. – Division Director Jules O’Rear, Ph.D. – Virology Team Leader Lisa Naeger, Ph.D. – Senior Virology Reviewer Damon Deming, Ph.D. – Virology Reviewer

OTS/Office of Computational Science Lilliam Rosario, Ph.D. – Office Director Charles Cooper, M.D. David Epstein, PMP

CBER/HIVE Team Carolyn Wilson, M.D. – Director Vahan Simonyan, Ph.D.

Office of the Chief Scientist Ross Felice, M.D. CDRH Scientific Computing Lab Brian Fitzgerald, Ph.D. – Director Mike Mikailov Fu-Jyh Luo

Office of Anti-Microbial Products Ed Cox, M.D. – Office Director David Roeder, M.S. – ADRA